"Power Optimizer" (solar) internals

[Curious.George]

Anyone ever take one of these apart to give an idea of what's inside?

I suspect they are MPPT controllers. But, some seem to have some "smarts" (to "talk" to the main inverter).

I'm not keen on destroying one just to have a look inside...

[r-p]

Re: "Power Optimizer" (solar) internals

We've had discussions on an electronics forum about the working, but no one has ever opened one (from the people who contributed).

Father-in-law has a SolarEdge system. These optimizers deliver 350V at all times. (If I understand correctly).
So the best I can come up with is that each module (lets say 10) knows what MPPT power it can achieve and communicates this through the powerwires to the inverter. This one adds up al this power, lets assume 1050W and devides this by 350V (which it wants to see). So the optimizers ALL have to supply 3A and the voltage to match (which may be 10V for a shadow panel and 50V for a panel in the full sun, just as long as it all adds up to 350V again.
This is then re-evaluated every few seconds. And reduce output to 1V if no communication is possible with the inverter.

So yes, a communication module, an MPPT tracker/calculator, a DC-DC converter, a microprocessor to manage it all and undoubtedly stuff I wasn't clever enough to imagine...

Keep us posted if you DO decide to open one up... (and you manage to get all the goo that is undoubtedly on there) from the components.

PS: supposedly the optimizers can also work WITHOUT a SolarEdge inverter. But it is beyond me how this would work.

[Curious.George]

Re: "Power Optimizer" (solar) internals

Father-in-law has a SolarEdge system. These optimizers deliver 350V at all times. (If I understand correctly).

The optimizers are mounted on individual panels. So, the string may try to deliver 350V to the inverter but a single panel won't (typ 40V). I suspect even that 350V is optimistic as if the entire array is shaded the entire string's output can drop (I think the inverters need ~150V from the string, at a minimum).

So the best I can come up with is that each module (lets say 10) knows what MPPT power it can achieve and communicates this through the powerwires to the inverter. This one adds up al this power, lets assume 1050W and devides this by 350V (which it wants to see). So the optimizers ALL have to supply 3A and the voltage to match (which may be 10V for a shadow panel and 50V for a panel in the full sun, just as long as it all adds up to 350V again.

But, therein lies the rub: just because the string can deliver a maximum of 1050W at some set of (V,I)'s, doesn't mean it can deliver 1050W at a shared value of I (cuz I must be constant for all panels) and with potentials that total "350V".

This is the step I can't fathom in the algorithm. I would assume the individual optimizers report a small amount of data. I.e., they likely don't report:

"Hi, I'm optimizer #6. I can give you 40V @ 7A or 38V @ 7.1A or 32V @ 6.5A or... What would you like??"
"Hmmm... wait a bit while I check to see what optimizer #7 can do for me..."

Rather, I imagine they report their present output (V,I) discovered from looking for the MPP on their panel at this time.

And, if "told" to deliver a specific current (seems the only logical way to ensure that each panel's output is compatible with its peers in the string), then how does it hunt for the MPP under those particular conditions? I.e., you've tied one of its hands...

This is then re-evaluated every few seconds. And reduce output to 1V if no communication is possible with the inverter.

But, my understanding is that the updates are required simply to adjust changing solar conditions (cloud cover, etc.). Or, do you think the inverter plays a second-order MPPT algorithm? For example:

"OK, let me tell the string to configure itself for 6.8A and see what sort of potential it develops... OK, now let me see if things improve when I configure it for 6.7A... 6.9A... Ooops! Maybe that was a cloud passing overhead..."

Keep us posted if you DO decide to open one up... (and you manage to get all the goo that is undoubtedly on there) from the components.

They're pretty pricey so I'd not want to "sacrifice" one unless I knew I would gain more knowledge than the value of having the actual devices "in my hands".

PS: supposedly the optimizers can also work WITHOUT a SolarEdge inverter. But it is beyond me how this would work.

That's where I wonder how much of the algorithm is "distributed" (among the optimizers) vs. centralized in the inverter.